1) Pixel -oriented visualization techniques

Pixel-Oriented Visualization Techniques are a type of data visualization method used to

represent large multidimensional datasets in a compact and efficient way by mapping
individual data values to colored pixels.

🔹 Key Characteristics:

1. Screen Division:
o For an m-dimensional dataset, the screen is divided into m windows, one for
each dimension.
2. Pixel Mapping:
o Each record in the dataset is visualized as a set of m colored pixels, where
the position of each pixel is fixed and corresponds to a dimension.
o The color of a pixel reflects the value of that dimension (e.g., using a color
scale from blue to red).
3. Space Efficiency:
o Highly space-efficient and suitable for displaying millions of data points on a
standard screen.
4. Pattern Detection:
o Enables the discovery of patterns, correlations, and anomalies across
dimensions.
5. Compact Visualization:
o To preserve space and show relationships, space-filling curves (e.g., Hilbert
curve or circular segment filling) may be used.

🔹 Advantages:

• Excellent for visualizing large datasets.

• Provides a qualitative overview of the data.
• Helps in detecting clusters, outliers, and data distribution.

🔹 Limitations:

• Interpretation may require domain knowledge or interaction tools (like brushing

and linking).
• Color mapping may be hard to distinguish if not well-designed.

2) Geometric Projection Visualization Techniques

• These techniques use geometric transformations and projections to represent
multidimensional data in lower dimensions (usually 2D or 3D), helping users detect
patterns and relationships.
Key methods include:

1. Scatter Plots & Matrices: Visualize relationships between 2 or more variables

using
• point plotting.
• ▪ It is one of the most commonly used plots used for visualizing simple data in
• Machine learning and Data Science.
• Useful for detecting correlation and clusters

2. Line plot:

• Connect data points with lines to show trends.

• ▪ The plot may seem very simple but it has more applications not only in machine
learning but in many other areas
• Commonly used for time series or performance metrics like ROC-AUC curves
 3. Bar/Stacked Bar Charts: Represent categorical data or part-to-whole relationships

• ▪ This plot may be simple and clear but it’s not much frequently used in Data
• science applications.
• Unlike a Multi-set Bar Graph which displays their bars side-by-side, Stacked Bar
Graphs segment their bars. Stacked Bar Graphs are used to show how a larger
category is divided into smaller categories and what the relationship of each part
has on the total amount

4. Box and Whisker Plots: Show data distribution, quartiles, and outliers.

• ▪ The straight lines at the maximum and minimum are also called whiskers.
• Points that lie outside the whiskers will
be considered as an outlier.
• The box plot also gives us a description
of the 25th, 50th,75th quartiles.
• With the help of a box plot, we can also
determine the
Interquartile range(IQR) where
maximum details of the data will be
present
 5. Pie & Donut Charts:Represent proportions and percentages.

• • A pie chart represents numbers in percentages, and the total sum of all
• Extensively used in presentations and offices, Pie Charts help show proportions and
percentages between categories, by dividing a circle into proportional segments

3) Icon-based visualization techniques

Icon-Based Visualization Techniques are used to represent multidimensional data using

small icons where different features of the icon represent different data attributes. This method
makes it easier to recognize patterns and relationships at a glance, even in high-dimensional data
sets.
🔹 Key Characteristics:

• Each data point is visualized as an icon

• Visual features (e.g., size, shape, angle, color) of the icon encode different
dimensions
• Particularly useful for comparative analysis and clustering
• Leverages human facial and shape recognition abilities

Common Techniques
1. Chernoff Faces

• Maps multiple dimensions to human facial features (e.g., eye size, nose length,
mouth curve)
• Takes advantage of humans' ability to detect slight differences in faces
• Useful for comparing multivariate data

Example:
A face may have:

• Big eyes for high values of variable A

• Curved mouth for positive values of variable B
• Narrow head for low values of variable C

2. Stick Figures

• Uses a human-like stick figure with limbs representing data

• Each limb’s angle/length maps to a different variable
• Two dimensions are often used for positioning; others are mapped to the figure’s
parts
 Example:

• Length of arms = income

• Angle of legs = education level
• Position on x-y plot = age and location

🔸 Advantages:

• Good for detecting clusters, anomalies, or trends in high-dimensional data

• Makes abstract data more intuitive

🔸 Limitations:

• May become hard to interpret with too many data points or dimensions

4) Hierarchical visualization techniques

Hierarchical Visualization Techniques represent data organized in tree-like structures where

elements have parent-child relationships. These techniques are especially useful when
visualizing data with multiple levels or nested categories. Based on your document, here are the
main hierarchical visualization techniques:

1. Treemaps

• Visualize hierarchical data using nested rectangles.

• Each rectangle represents a category and its size shows quantity (e.g., file size).
• Subcategories are nested within parent rectangles.
 • Useful for space-efficient overviews.
• Downside: Harder to perceive hierarchy depth compared to other techniques.

2. Circle Packing

• Similar to treemaps, but uses nested circles instead of rectangles.

• Shows containment well (each circle is within a parent circle).
• Less space-efficient but better at visualizing hierarchy structure.

3. Sunburst Diagrams

• Also known as Radial Treemaps or Multi-level Pie Charts.

• Root node is at the center, and levels expand outward as rings.
• Each arc or slice represents a category.
• The angle and size of slices show proportions and hierarchy.
• Good at showing both structure and proportions clearly.

4. Worlds-within-Worlds (n-Vision)

• A technique to visualize high-dimensional data in layers.

• Fix some dimensions to constant values to visualize how others behave.
• Especially helpful when dealing with 6-D or higher datasets.

Summary Table:
Space Hierarchy
Technique Shape Use Case
Efficiency Clarity

Visualizing files, storage,

Treemap Rectangles High Moderate
budgets

Display nested categories

Circle Packing Circles Moderate High
visually

Sunburst Multilevel proportions (e.g.,

Rings Moderate High
Diagram org charts)

Worlds-within- 3D Exploring multi-dimensional

Low High
Worlds Layering scientific data

5) Visualization complex data and relations

Visualizing Complex Data and Relations involves techniques to represent high-dimensional,
structured, or unstructured data in a way that human eyes and minds can perceive patterns,
correlations, and insights effectively. Here’s a breakdown of key techniques from the document:

🌐 1. Hierarchical Visualization of Subspaces

• Challenge: Visualizing all dimensions of high-dimensional data at once is

infeasible.
• Solution: Partition dimensions into subsets (subspaces) and visualize them
hierarchically.
• Example Technique:
o Worlds-within-Worlds (n-Vision): Fix certain dimensions (e.g., X3, X4, X5)
and observe the behavior of another (e.g., F) with respect to two others (e.g.,
X1, X2).

🔤 2. Word Cloud (Tag Cloud)

• Use case: Primarily for textual or categorical data.

• Features:
o Size represents frequency or importance.
o Color may denote categories or extra metadata.
• Limitations: Aesthetic, not analytically precise (e.g., long words appear more
prominent regardless of importance).

📊 3. Advanced Multidimensional Techniques

These techniques may be combined with other visualization styles (e.g., geometric or icon-
based) to handle complex relationships:

• Parallel Coordinates: Great for analyzing trends across many variables.

• Scatterplot Matrices & Hyperslice: Break down multidimensional data into
pairwise comparisons or slices.
• Chernoff Faces / Stick Figures: Represent multiple variables as features of an
icon.

📌 Use Cases

• Numeric data: Scatter plots, parallel coordinates, and n-Vision.

• Categorical/text data: Word clouds, tag clouds, mosaic (Marimekko) plots.